Medical Imaging Change Detection via Image-Text Feature Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Interpreting follow-up medical images to assess changes compared to baseline images is time-consuming and challenging due to artifacts from differing imaging parameters and irrelevant changes.
Innovation Solution
A computer-implemented framework that uses a trained image processing machine learning model to generate an image feature vector representing the difference between first and second medical imaging data. This vector is then compared to text feature vectors describing potential changes, selecting the most similar text feature vector to determine the data representing the change.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a radiologist manually reviews and compares baseline and follow-up medical images to assess condition progression, then accurate interpretation of medical changes can be achieved, but the process becomes time-consuming and labor-intensive
Solution Approach 1:
The patent introduces an intermediary system comprising image processing circuitry and natural language generation circuitry that acts as a mediator between the medical images and the radiologist. The image processing circuitry automatically compares baseline and follow-up images to identify changes, while the natural language generation circuitry creates descriptive reports of these changes. This intermediary system handles the time-consuming comparison and documentation tasks, allowing the radiologist to focus on final assessment and decision-making, thus reducing review time while maintaining interpretation accuracy.
2Adaptability or versatility
If baseline and follow-up images are captured using different imaging parameters (lighting, angle, etc.), then flexibility in image acquisition is improved, but artifacts are introduced that complicate image interpretation
Solution Approach 1:
The patent applies parameter changes by using image processing techniques that normalize and adjust for variations in imaging parameters between baseline and follow-up images. The system processes images to compensate for differences in lighting, angle, and other acquisition parameters, transforming them into a common reference frame. This allows the system to maintain the flexibility of using different imaging parameters during acquisition while eliminating their confounding effects during comparison, thereby reducing interpretation difficulty without sacrificing acquisition adaptability.
3Productivity
If automated image comparison algorithms are used to reduce interpretation time, then processing speed is improved, but accuracy in identifying relevant changes may deteriorate due to false positives from artifacts
Solution Approach 1:
The patent implements feedback mechanisms where the automated image comparison system continuously refines its analysis based on the generated descriptions and identified changes. The system processes images to detect changes, generates natural language descriptions of these changes, and uses this feedback to improve subsequent comparisons. This iterative feedback loop allows the system to learn from its own outputs, distinguish between relevant medical changes and artifacts more effectively, and improve detection accuracy while maintaining high processing speed.
Data Source
AI summary
A computer implemented framework for determining data representing a change between first medical imaging data and second medical imaging data is disclosed. Imaging data representative of a difference between the first medical imaging data and the second medical imaging data is obtained. The imaging data is input into a trained image processing machine learning model to generate an image feature vector. A plurality of text feature vectors is obtained, each text feature vector being representative of natural language text describing a respective change in medical imaging data. For each of the plurality of text feature vectors, a similarity measure indicating a degree of similarity to the image feature vector determined. A text feature vector is selected. The data representing the change between the first medical imaging data and the second medical imaging data is determined based on the selected text feature vector.